Turnover at the Top: Investigating Performance-Turnover Sensitivity among Nonprofit Organizations

ABSTRACT

Research has confirmed private firm performance as a predictor of executive turnover, but whether this relationship holds in other organizational contexts—especially those operating without a profit-distribution requirement—is not known. For nonprofit organizations, performance monitoring and governance largely occur behind closed doors under the leadership of a volunteer board of directors. Nonprofits are commonly criticized as inefficient, even ineffective, and the accountability of boards and executives for organizational performance has not been sufficiently investigated. This article engages a unique dataset of 998 U.S.-based nonprofits serving missions related to the arts, health, and human services, and uses a multiple-spell discrete-time hazard model to evaluate nonprofit financial performance and the likelihood of executive turnover. The findings provide preliminary support that tenure in the executive office is sensitive to nonprofit financial performance, and in doing so, they raise new insights about the accountability of nonprofit executives.

KEYWORDS:

• executive behavior
• nonprofit management
• performance management

Acknowledgment

The authors would like to thank the anonymous reviewers whose feedback helped strengthen the manuscript.

Notes

Research on financial reporting of nonprofit organizations has documented the limitations to research that are posed by the reliability of nonprofit financial data (Tinkelman & Mankaney, 2007; Trussel & Parsons, 2007). Following Calabrese (2011), the sample was limited to exclude annual observations that included “obviously” erroneous financial data, including negative assets, liabilities, revenues, or expenses. In total, 1,617 year observations were dropped from the sample. Finally, to create a panel that included observations for each organization before, during, and after a turnover event, observations were grouped according to organization and fiscal year to ensure that each organization had at least one observation for a pre-time period (i.e., 2005 or 2006), during-time period (i.e., 2007, 2008, or 2009), and post-time period (i.e., 2010 or 2011).

These mission markets, along with their NTEE codes, included: Museums (A50–A57); Performing arts (A62–A6C); Community health treatment (E30–E42); Abuse prevention (I70–I73); Employment and vocational training (J20–J33); Nursing, home health care (E90–E92); Substance abuse prevention and treatment (F20–F22); Hotlines and crisis prevention (F40–F42); Crime prevention and rehabilitation (I20–I44); Food pantries and programs (K30–K36); Public housing and rehabilitation (L21–L25); Homeless shelters (L40–L41 & P85); Community centers (P28); Family counseling (P46); Senior centers (P81); Residential care and group homes (P73).

These nonprofits had insufficient information listed in GuideStar for one or more of the following reasons: no information for organization in GuideStar, no executive listed, not enough information to identify turnover, or executive label is unclear in GuideStar data.

The “margins” in Stata were used to convert the log odds estimate to a probability.

Of course, the same is true of all covariates in a nonlinear model. For excellent discussions of these issues, please see Ai and Norton (2003), Karaca-Mandic, Norton, and Dowd (2012), Long and Freese (2014), Norton, Wang, and Ai (2004).

This is easily accomplished using the suite of “margin” commands in more recent versions of Stata.

Additional information

Notes on contributors

Amanda J. Stewart

Amanda J. Stewart is an Assistant Professor at North Carolina State University.

Jeffrey Diebold

Jeffrey Diebold is an Assistant Professor at North Carolina State University.